It's probably a myth.
"There are a ton of misconceptions about tubes including the belief that they sound 'warm.' Tube circuits have about as many sounds as there are circuits. It is more about the topologies used and aims of the designer as to how a circuit will sound. In fact, it is usually the transformers that are typically used in tube-based products that create the illusion of 'warmth.' There is a lot of vintage solid-state gear that used transformers and is commonly described as 'warm.' A typical transformer adds some odd-order distortion at the lowest frequencies and tends to restrict the ultrasonic frequencies, which we subtly perceive because of the phase shift. Part of this effect also seems to smooth transients. Tubes probably tend to offset some of that effect, because a little THD tends to exaggerate transients." -- Hutch Hutchison of Manley Labs.
"There are a ton of misconceptions about tubes including the belief that they sound 'warm.' Tube circuits have about as many sounds as there are circuits. It is more about the topologies used and aims of the designer as to how a circuit will sound. In fact, it is usually the transformers that are typically used in tube-based products that create the illusion of 'warmth.' There is a lot of vintage solid-state gear that used transformers and is commonly described as 'warm.' A typical transformer adds some odd-order distortion at the lowest frequencies and tends to restrict the ultrasonic frequencies, which we subtly perceive because of the phase shift. Part of this effect also seems to smooth transients. Tubes probably tend to offset some of that effect, because a little THD tends to exaggerate transients." -- Hutch Hutchison of Manley Labs.
Undoubtly, tube rectified supply leaves a sonic footprint in tube circuits, as does a regulated SS supply. Who like what and what betters the other? I think that's like asking wich wine is the best, peoples opinions will differ vastly, but most tend to prefer valve supply over regulated SS, and that is my personal experience as well. I use SLA batteries and like it better than SS or tube rectified supply. Charging can be a hustle though.
So myth or truth? I'd say both.
So myth or truth? I'd say both.
On my salary he couldn't!
Part of the character imposed on an amplifier's sound by a vacuum rectifier is due to sag, when the current demand increases causing the B+ voltage to drop. You can simulate this, to some extent at least, with an SS rectifier and some added series resistance.
Probably the best property of vacuum rectifiers, though, is the lack of "hash", which can cause noise with some SS rectifiers - not with Schottky diodes, though.
I must agree with posts 2 and 6.
Whatever is subjectively heard (and there could be hidden reasons for experiences, even as slight as someone entering the room or a slightly changed listening position), there is no scientific reason for the source of d.c. being able to make a difference. (And I hope that on a site the level of DIY-Audio we have moved away from snake oil.)
That is not to say that effects are not experienced subjectively. My measurements (scope on the h.t. rail) mostly support Ray_Moth's explanation, although this was more evident with older directly heated rectifiers and low filter capacitor values. There can also be an almost unperceivable effect with ss rectifier switching noise, especially at high voltage supplies, but cures for that are easy. I would certainly not opt for a tube rectifier with the associated heat and filament supply effects (and cost!) in preference to ss, then with a small serie resistor if you want to be pedantic. As for the initial voltage surge with the latter, that can also be taken care of.
Regards
Whatever is subjectively heard (and there could be hidden reasons for experiences, even as slight as someone entering the room or a slightly changed listening position), there is no scientific reason for the source of d.c. being able to make a difference. (And I hope that on a site the level of DIY-Audio we have moved away from snake oil.)
That is not to say that effects are not experienced subjectively. My measurements (scope on the h.t. rail) mostly support Ray_Moth's explanation, although this was more evident with older directly heated rectifiers and low filter capacitor values. There can also be an almost unperceivable effect with ss rectifier switching noise, especially at high voltage supplies, but cures for that are easy. I would certainly not opt for a tube rectifier with the associated heat and filament supply effects (and cost!) in preference to ss, then with a small serie resistor if you want to be pedantic. As for the initial voltage surge with the latter, that can also be taken care of.
Regards
I should have added that I personally know nothing about these things. But I do like the quote. Especially because it supports what I LIKE to think. And I LIKE to think it's about engineering and not technology. Because if it is the latter, then what hope do we have? What use is there for engineering, for competence, if all it takes is a tube?
I too think "hash" could be a problem. But isn't that something mostly associated with early/ier ss? I will not force this point. It's just a thought.
I too think "hash" could be a problem. But isn't that something mostly associated with early/ier ss? I will not force this point. It's just a thought.
Johan Potgieter said:
It suddenly dawns on me that every time i replace the vacuum rectifiers with mercury my cat enters the room. All i hear must be due to the additional absorption.
I also wonder if wearing some headgear tuned to cuttoff everything above 7khz will help me with my delusions.
Johan Potgieter said:
I must agreeably disagree with Post 7.
I wish I could remember which project it was, probably the single stage SE headphone amp based on a 6C45 driving a 5K OPT and a 6X-something rectifier, but I was more than a bit surprised to see the PS ripple looked like two superimposed sine waves instead of the ubiquitous triangle typical of SS supplies. Almost all common circuits have at least one SE stage, typically the most sensitive front end, it's not that adventurous to suggest wide band harmonics on the PS could contribute to a device's sonic signature.So many variables intervene: RC or CLCLC or LCLC? Power amp or pre? High drop tube rectifiers or damper diodes like the 6D22S? SE or PP? Is the system in a high RF industrial area or out on the farm? It's hard to see how a single technique could be the best solution for them all. That said, some vere highly respected 'tweak' firms go exclusively SS, like Manley.
I've heard the difference between a 5U4G and a 5U4GB in a PP 6V6 amplifier.
The 5U4G had slightly softer bass,and slightly smoother mids/highs,and the 5U4GB had a nice overall balance,bass was better,and mids/highs were close.I settled on the 5U4GB.
I'm not sure if it was so much the rectifers,or the slightly differing B+ voltage between the two tubes,but I could notice it.
The 5U4G had slightly softer bass,and slightly smoother mids/highs,and the 5U4GB had a nice overall balance,bass was better,and mids/highs were close.I settled on the 5U4GB.
I'm not sure if it was so much the rectifers,or the slightly differing B+ voltage between the two tubes,but I could notice it.
OK, analog_sa, poke some local fun. My ears may be suffering from what I hear about our present cricket and rugby.
But my remark regarding room conditions was not totally flippant. A person standing close to a loudspeaker was actually shown to have made a slight difference (cutting h.f. propagation to a reflecting surface), and tests I once did showed a difference in response graphs that would have been audible, when moving the microphone the width of one person.
Rdf, I totally agree - my remarks should have said "all other things being equal". Although one usually finds the most sensitive stage (as you correctly pointed out) at the input, but then having been decoupled by at least one RC, which in my experience usually filtered ripple to an almost non-existing degree. As said, ss diode switching transients can be eliminated, and I would be surprised to find early stages so poorly filtered that power supply ripple will interfere. But to agree with you, in the complete picture it is a valid factor if not taken care of. I still think that h.t. sag is the primary factor here in circuits otherwise well-designed.
Regards.
But my remark regarding room conditions was not totally flippant. A person standing close to a loudspeaker was actually shown to have made a slight difference (cutting h.f. propagation to a reflecting surface), and tests I once did showed a difference in response graphs that would have been audible, when moving the microphone the width of one person.
Rdf, I totally agree - my remarks should have said "all other things being equal". Although one usually finds the most sensitive stage (as you correctly pointed out) at the input, but then having been decoupled by at least one RC, which in my experience usually filtered ripple to an almost non-existing degree. As said, ss diode switching transients can be eliminated, and I would be surprised to find early stages so poorly filtered that power supply ripple will interfere. But to agree with you, in the complete picture it is a valid factor if not taken care of. I still think that h.t. sag is the primary factor here in circuits otherwise well-designed.
Regards.
SY said:
In a properly designed circuit the resistor won't glow? Sorry, couldn't resist. I guess the real question would be if the result was a better circuit. It's natural to think more of a desired parameter (phrased clumsily) is always better. More PS capacitance is better, lower PS impedance is always better, lower PS transformer losses are always better, etc.. The light really went on for me watching a thread over in the ClassD forum. Bruno was cautioning against a wave of modders wanting to 'upgrade' the PS caps on his digital amp modules. The stock caps were chosen to have an optimum, not minumum, amount of internal resistance. Changing them invited disturbing the circuit balance at high frequencies. Another example was a discussion around RC filtering in a tube phono or pre supply. A convincing case was made that the HF filtering of a cascade of RC pairs far outweighted reduced regulation of a lightly loaded supply.
I don't have any experience comparing a diode/resistor to a tube rectifier but don't see any reason the same wouldn't hold here. Maybe the D/R is a better solution than D alone. Returning to the original question of tube sound and tube rectifiers, it could well be the latter contributes for positive technical reasons in some cases, and is detrimental in others. It might also explain why some of the lowest loss rectifiers, like a 5ar4, have the poorest sonic reputations.
rdf said:
Hi rdf,
Please allow me to respectfully disagree with your speculations. Any attempt to use Ohm's law as a sole explanation for the more subtle sonic differences seems futile. I also dislike the 5ar4 but to blame low impedance seems too simplistic. What about the (universally considered?) great sounding Mercury and tv dampers rectifiers? They have losses comparable to SS.
And can anyone explain or at least speculate, why high voltage (cree) Schottkys seem so good an paper, yet so few like them?
No need to stand on ceremony, 'de-bunkify' with prejudice where warranted. I have a trio of new Cetron 866's awaiting a suitable project to learn about mercury vapour rectifiers down the road. You may be entirely right. The discussion had drifited toward sag being the primary mechanism at work and my thought was that a supply designed for low DC supply impedance above all else might not result in the best supply in all audio applications. It's difficult to see how from a purely engineering perspective that it's warranted for high gain, low level, relatively constant current devices such as single-ended pre-amps, phono pre's, etc., when some R at the rectifier lessens the potential of generating RF noise.
I took a quick look at the Cree site, aren't those diodes ultra-fast 'hard' recovery? I thought the preference had shifted to soft recovery diodes to minimize generation of RF hash.
I took a quick look at the Cree site, aren't those diodes ultra-fast 'hard' recovery? I thought the preference had shifted to soft recovery diodes to minimize generation of RF hash.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.